GF3 should ramp much faster:
1. Musk isn't involved (I sometimes joke he's barred from the premises, haha)
2. We already know the parts all fit together
The main risks:
1. Design "improvements" using new parts that don't fit together on the first try
2. Parts sourced from different vendors that have slight differences
3. Equipment from different vendors that makes parts with slight differences
I think these risks are low, especially at first when most parts will come from the same vendors or be made in Tesla's US factories using the same Tesla machines. They can substitute local production gradually, testing each part or subassembly without holding up the rest of the line.